It’s a simple concept that sits at the core of representative democracy: The party that wins the most votes should win the most seats. It’s practiced in free countries around the world, to ensure, as a matter of basic fairness, that those in power have the consent of those they govern and are responsive to changes in the electorate.
But not, ironically, in the nation considered the birthplace of modern democracy—the United States of America.
The Electoral College, of course, allows for the perverse possibility that the candidate who earns the most votes will not win the presidency, a fate that’s befallen us twice now in the last two decades alone. And the Senate thumbs its nose at majority rule by design, giving voters in the smallest state 68 times the political power of those in the largest.
The House is supposed to be the counterweight to these retrograde institutions, the one body that’s intended to represent the people as closely as possible—it even has the word “Representatives” in its name. But it, too, is badly flawed. Thanks in large part (but not solely) to gerrymandering, there’s simply no guarantee that control of the House will align with the party preference of the voters.
In fact, Democrats could win the national popular vote for the House by a wide margin in November yet still fall short of a majority of seats. Below we’ll look at how and why this is possible, and what election-watchers should expect based on the data available to us.
While control of the House plays out in the 435 separate districts that make it up, it’s impractical to poll so many races. To get an approximation of the complete picture, therefore, pollsters ask a nationwide audience which party they plan to vote for in the House. This is known as the generic congressional ballot, and currently, Democrats hold a 9-point lead according to averages of public polls compiled both by HuffPost Pollster and FiveThirtyEight. (In general, these polls do a good job of predicting the actual outcome of the national House vote.)
If this 9-point spread holds through Election Day, it’d be larger than the 7-point margin by which Republicans won the House vote in their 2010 landslide, when they picked up 63 seats and, with them, took charge of the chamber. It would also be bigger than the margin in 1994, when Republicans also won by 7 points and gained 54 seats, and even 2006, when Democrats won by 8 points and reclaimed the majority by flipping 31 seats. In fact, in the last 30 years, the only more impressive performance came in 2008, a year in which Democrats romped to a 10-point national win and added 21 seats to their caucus.
Yet even though voters favor Democrats by near-historic proportions this year, there’s still no guarantee that the party will flip the 23 seats it needs to gain control of the House. Indeed, in 2012, Democratic candidates actually did win more votes than Republicans, yet the GOP still held the House. So why is the House stacked against Democrats in such an anti-(small-d)-democratic way?
Because we elect Congress via single-member districts, the way those districts are drawn plays an enormous role in shaping the results. Furthermore, it gives rise to the pernicious problem of gerrymandering, where mapmakers can give one party an advantage by packing the other party into a minority of seats that they win by lopsided margins while making sure the favored party wins a majority of seats by narrower—yet still safe—margins.
As a result of the GOP's 2010 landslide, which gave them control over most of the state governments responsible for this decade's redistricting, states with five times more congressional districts were drawn to favor Republicans over Democrats. You can see this starkly in the map below.
The district lines for 2018 saw Donald Trump win 228 congressional districts to just 207 for Hillary Clinton, even though Trump lost the popular vote by 2.1 points nationwide. You can also see this illustrated in the next graphic, which is known as a histogram. This chart buckets House districts by their 2016 presidential margin, and it shows that the seats Trump won are clustered in a zone where many are just red enough to ensure Republicans will hold them. The Clinton seats, by contrast, stretch much farther to the left, revealing how Democratic voters have in many instances been packed into a handful of ultra-blue seats.
The chart demonstrates something else as well. The district in the very middle, Nebraska’s 2nd, backed Trump by 2.2 points in 2016. Given Clinton’s 2-point national margin, that means this median district was over 4 points to the right of the country as a whole. In other words, to secure a bare majority of 218 seats in the House, Republicans, at least in theory, would not have to win any blue seats. Democrats, however, would have to win a number of red seats—11, in fact.
This phenomenon makes it extremely hard for Democrats to translate their share of the national vote into an equivalent share of seats. It’s usually very difficult for one party to win congressional districts that were carried by the other party’s presidential candidate. You can see this illustrated in the next chart, which sorts House seats into four quadrants. The vast majority fall into the bottom-left corner (districts won by Trump and held by Republicans) or the top right (those won by Clinton and held by Democrats).
Just a tiny number of Clinton seats were won by Republicans in 2016—23 in total—and even fewer Trump seats were carried by Democrats: only 12.
That’s due to historically high polarization between the parties (meaning there’s virtually no ideological overlap), historically low levels of ticket-splitting (in other words, few people are willing to vote for a Republican president and a Democratic member of Congress, or vice-versa), and historically low advantages for incumbents (who traditionally have been better-able to convince voters to split their tickets by virtue of being more familiar to the electorate and having a record of accomplishments to point to). As a result, the correlation between presidential results and House results are at near-record levels.
However, these correlations aren’t perfect, meaning there's some amount of variance between presidential and congressional outcomes. And that’s more likely to be true in a midterm election like 2018, when Trump won’t be on the ballot. Adding further to the uncertainty is that, while incumbency bonuses are at historic lows, they do still exist.
As a result, there’s disagreement about just how big of a popular vote margin Democrats would need nationally to enable them to win a majority in the House. But the news isn’t all bad, since there are three important factors that weigh in favor of Democrats this year. However, since these considerations didn’t pertain to election cycles earlier this decade, they complicate the picture further.
First on the positive side of the ledger, Pennsylvania's Supreme Court threw out the GOP's gerrymander earlier this year and replaced it with a fairer map. Second, a record number of Republicans decided not to seek re-election in 2018. That’s led to an all-time high of 41 open GOP seats (as shown below), which has eroded the party’s incumbency advantage.
Finally, Democratic candidates have shattered fundraising records this year, and not just in their top-tier of offensive opportunities, but in scores of seats. In the third quarter of 2018, Democrats outraised their Republican opponents in 93 of the 100 races that Daily Kos Elections had rated as potentially competitive for either party when fundraising reports came out in mid-October, often by several times over. Candidates with a financial advantage don’t prevail in every race, but having the funding necessary to run a real campaign and get your message out is critical in House races, where challengers typically start off with much lower name recognition than incumbents.
Lastly, there's one further variable that can distort the House popular vote itself compared to national pre-election polls: seats where one major party isn't on the ballot, since poll respondents may favor one party only to find they can't actually vote for them. Republicans don't have a candidate in 39 districts, but Democrats have one on the ballot in all but three districts, their fewest uncontested seats since the Watergate election of 1974 and the biggest gulf between the parties since the 2006 Democratic wave.
Such a lopsided disparity means that millions of voters who tell pollsters they would vote Republican won't actually get the chance to do so, even if they almost certainly would have if given the opportunity. By contrast, very few Democratic voters will find themselves in the same situation. And note that in almost all states, uncontested races will still see votes counted for the one major party that’s still on the ballot.
This gap will therefore skew the 2018 popular vote in favor of Democrats compared to what the generic ballot polls predict. In other words, if the polls are accurate, they will almost certainly underestimate the Democrats’ ultimate national vote margin simply because so many fewer Republicans are on the ballot, but by exactly how much, it’s difficult to say.
Yet despite all the uncertainty, several analysts and pollsters have nevertheless tried to forecast just how big a win in the national vote Democrats would need to also win a majority of seats in the House.
On the high end of the range is a report from New York University’s Brennan Center for Justice in which redistricting law experts estimate that Democrats would have to win the national popular vote by a huge 11 points to have an even shot at taking a majority. This argument has support from the fact that Republicans won the middle-most House district—that is, district No. 218—by 13 points in 2016 even though they won the national vote by just 1 point, suggesting the GOP had a 12-point advantage.
If, therefore, every district swung toward Democrats by an equal amount in 2018 compared to 2016, it would take a shift of approximately the size that the Brennan Center proposes in order for Democrats to capture a majority. This conceit, known as a “uniform swing,” of course can never play out quite so perfectly in reality, but it’s a useful conceptual approach. However, the Brennan Center’s estimate doesn’t factor in the GOP’s reduced incumbency advantage in 2018, which will likely make the swing away from 2016’s results even less uniform than it otherwise may have been.
Another estimate comes from FiveThirtyEight, which offers multiple versions of their House prediction model. One model that takes into account just local and national polls estimates Democrats need a 7.2-point margin for a House majority. But another version that adds in the “fundamentals” such as fundraising, past voting history, and historical national trends lowers that margin to a less-daunting 5.5 points. FiveThirtyEight currently gives Democrats strong odds of managing either feat and taking back the House.
The pollster YouGov produced a very similar estimate back in June saying that Democrats would need a 5.4-point margin to eke out a narrow majority of just 219 seats. (More recently, YouGov has predicted a somewhat wider win for Democrats.)
Meanwhile, the Economist’s Elliott Morris has his own model that relies on both district-specific polling and national polling, along with past election results. It then runs a simulation to produce a range of possible outcomes. Morris estimates that Democrats would need roughly a 6-point national margin to flip the House, and he also thinks they’re favored to win the House.
Another estimate comes from Rachel Bitecofer of Virginia's Christopher Newport University, who relies on presidential results by district and incumbency, plus demographics such as education, race, and urbanization. Bitecofer agrees that Democrats probably need at least a 6- to 8-point margin, but her model predicts Democrats take the House with room to spare in the present environment.
Finally, one of the lowest estimates comes from Emory University political scientist Alan Abramowitz, who projects that Democrats only need a 4-point margin to take the House. However, this model is largely based on historical post-World War II elections and doesn’t account for important factors like the decline of ticket-splitting, the increase in partisan polarization, and the much stronger impact of gerrymandering in 2018 compared to previous cycles. It also doesn’t address the circumstances regarding individual races and candidates.
So what to make of all these differing estimates, which range from 4 points all the way to 11? One of the key lessons is that there’s simply a lot of uncertainty about how big a margin Democrats need to flip the House. Based on Daily Kos Elections' own assessment of the GOP's advantage—from gerrymandering and single-member districts, plus the importance of polarization, incumbency or the lack thereof, and the strong correlation between presidential and House results—we conclude Democrats may need a 6- to-7-point national margin to flip the House, but if a higher or lower outcome proves necessary, that shouldn’t be a surprise.
There’s something even more important, though: The size of the House popular vote just shouldn’t have to be a concern. In a true democratic republic, especially one with only two parties, any legitimate electoral system will ensure that whichever party wins the most votes also wins the most seats. If Democrats—and the country—are fortunate in November, that’s what will happen, but there’s a real chance it won’t.